Towards Safe Reinforcement Learning via Constraining Conditional Value-at-Risk

TL;DR

This paper introduces a risk-sensitive reinforcement learning method called CPPO that constrains the Conditional Value-at-Risk (CVaR) to improve safety and robustness against uncertainties in transition and observation, validated on MuJoCo tasks.

Contribution

It provides a theoretical analysis linking performance degradation to a new metric VFR and proposes a novel CVaR-based constrained optimization algorithm for safer RL.

Findings

CPPO achieves higher cumulative rewards.

CPPO demonstrates robustness against disturbances.

Theoretical link between VFR and performance degradation.

Abstract

Though deep reinforcement learning (DRL) has obtained substantial success, it may encounter catastrophic failures due to the intrinsic uncertainty of both transition and observation. Most of the existing methods for safe reinforcement learning can only handle transition disturbance or observation disturbance since these two kinds of disturbance affect different parts of the agent; besides, the popular worst-case return may lead to overly pessimistic policies. To address these issues, we first theoretically prove that the performance degradation under transition disturbance and observation disturbance depends on a novel metric of Value Function Range (VFR), which corresponds to the gap in the value function between the best state and the worst state. Based on the analysis, we adopt conditional value-at-risk (CVaR) as an assessment of risk and propose a novel reinforcement learning algorithm of CVaR-Proximal-Policy-Optimization (CPPO) which formalizes the risk-sensitive constrained optimization problem by keeping its CVaR under a given threshold. Experimental results show that CPPO achieves a higher cumulative reward and is more robust against both observation and transition disturbances on a series of continuous control tasks in MuJoCo.